Systematic review of targeted axillary dissection in node-positive breast cancer treated with neoadjuvant systemic therapy: variation in type of marker and timing of placement

Abstract

Background

In node-positive (cN+) breast cancer treated with neoadjuvant systemic therapy, combining sentinel lymph node biopsy and targeted lymph node excision, that is targeted axillary dissection, increases accuracy. Targeted axillary dissection procedures differ in terms of the targeted lymph node excision technique. This systematic review aimed to provide an overview of targeted axillary dissection procedures regarding definitive marker type and timing of placement: before neoadjuvant systemic therapy (1-step procedure) or after neoadjuvant systemic therapy adjacent to a clip placed before the neoadjuvant therapy (2-step procedure).

Methods

PubMed and Embase were searched, to 4 July 2023, for RCTs, cohort studies, and case–control studies with at least 25 patients. Studies of targeted lymph node excision only (without sentinel lymph node biopsy), or where intraoperative localization of the targeted lymph node was not attempted, were excluded. For qualitative synthesis, studies were grouped by definitive marker and timing of placement. The targeted lymph node identification rate was reported. Study quality was assessed using a National Institutes of Health quality assessment tool.

Results

Of 277 unique records, 51 studies with a total of 4512 patients were included. Six definitive markers were identified: wire, ¹²⁵I-labelled seed, ^99mTc, (electro)magnetic/radiofrequency markers, black ink, and a clip. Fifteen studies evaluated one-step procedures, with the identification rate of the targeted lymph node at surgery varying from 8 of 13 to 47 of 47. Forty-one studies evaluated two-step procedures, with the identification rate of the clipped targeted lymph node on imaging after neoadjuvant systemic therapy varying from 49 to 100%, and the identification rate of the targeted lymph node at surgery from 17 of 24 to 100%. Most studies (40 of 51) were rated as being of fair quality.

Conclusion

Various targeted axillary dissection procedures are used in clinical practice. Owing to study heterogeneity, the optimal targeted lymph node excision technique in terms of identification rate and feasibility could not be determined. Two-step procedures are at risk of not identifying the clipped targeted lymph node on imaging after neoadjuvant systemic therapy.

In node-positive (cN+) breast cancer treated with neoadjuvant systemic therapy, combining sentinel lymph node biopsy and targeted lymph node (TLN) excision (targeted axillary dissection, TAD) increases accuracy. TAD procedures differ in terms of the TLN excision technique. This systematic review aimed to provide an overview of TAD procedures regarding definitive marker type and timing of placement. Owing to study heterogeneity, the optimal technique could not be determined. High-quality prospective studies reporting all relevant aspects of TAD are needed.

Introduction

In clinically node-positive (cN+) breast cancer, axillary lymph node dissection (ALND) is associated with substantial morbidity^1,2, but used to be standard of care. At present, patients with cN+ disease often receive neoadjuvant systemic therapy (NST). After NST, approximately one-third of patients achieve an axillary pCR^3–6, which is associated with improved prognosis compared with having residual axillary disease^7–10. Less invasive axillary staging procedures were therefore proposed in an effort to enable response-guided treatment, by identifying an axillary pCR so that ALND could be omitted in such patients. Currently, several less invasive axillary staging procedures are being performed worldwide.

Several studies have assessed the diagnostic accuracy of these less invasive axillary staging procedures compared with ALND in patients with cN+ disease. Trials^6,11–13 such as SENTINA, SN FNAC, and ACOZOG Z1071 have shown that performing sentinel lymph node biopsy (SLNB) after NST results in false-negative rates (FNRs) of 14.2, 13.3, and 12.6% respectively, and a negative predictive value (NPV) that does not exceed 86%. Using dual tracers, immunohistochemistry, and excising at least three sentinel lymph nodes (SLNs) can improve the FNR⁶. The median number of SLNs detected is two¹⁴, and so recommending removal of three or more SLNs may result in node-picking, whereby non-SLNs are also removed. An alternative to SLNB is to specifically target a metastatic axillary lymph node by placing a marker inside it before NST. After NST, this targeted lymph node (TLN) is localized using visual inspection, imaging, or probe-guided methods, and subsequently excised. For example, when the marking the axilla with radioactive iodine (MARI) procedure is undertaken¹⁵, a radioactive iodine-labelled seed (¹²⁵I seed) is placed before NST, followed by excision of the TLN after NST under the guidance of a hand-held γ probe. The MARI procedure, first described in 2010¹⁶, has an FNR of 7% and NPV of 83.3%. This is comparable to the NPV of SLNB. Lastly, SLNB and excision of a TLN can be combined in the procedure called targeted axillary dissection (TAD)¹⁷.

In a subanalysis of the Z1071 trial¹⁸, published in 2016, a clip was placed in a metastatic axillary lymph node before NST in 170 patients. Intraoperative localization of the clipped lymph node was not attempted, yet reporting whether it was located in either the SLNB or ALND specimen was encouraged. In 29 of 170 patients (24.1%), the clipped lymph node was reported to be found in the ALND specimen, suggesting that performing TAD improves diagnostic accuracy by removing additional relevant lymph nodes¹⁸. Three studies^17,19,20 assessing TAD in 35–85 patients reported an FNR that varied from 2 to 4%, and an NPV that ranged from 92 to 97%. In 2022, a Dutch prospective multicentre trial²¹ investigating radioactive iodine seed localization in the axilla with the sentinel node procedure reported an FNR of 3.5% and an NPV of 92.8% among 212 patients, confirming the superior diagnostic accuracy of TAD. Studies of oncological outcomes, and especially impact on quality of life, of response-guided axillary treatment based on less invasive axillary staging techniques remain limited^22–24.

Meanwhile, a wide variety of TAD procedures are being incorporated into clinical practice, with variation in the type of definitive marker used (for example, magnetic marker, black ink, wire, clip)^20,25–27, as well as the timing of definitive marker placement (before or after NST). The technique used may affect ability to identify the TLN. The aim of this systematic review was to provide an overview of studies reporting on TAD in cN+ breast cancer treated with NST, focusing on types of marker used for TLN excision, timing of marker placement, and ability to identify the TLN.

Methods

Inclusion criteria

The PRISMA checklist was used for this systematic review²⁸. A systematic literature search was made for RCTs, cohort studies, and case–control studies with a minimum of 25 included patients describing experience with TAD in cN+ breast cancer treated with NST. Study protocols, conference abstracts, case reports, editorials, commentaries, and reviews were excluded, as were studies for which the full text was not available in English. Pathological confirmation of nodal positivity was not required, as the focus was on the surgical technique and the identification rate (IR) of the TLN, rather than on diagnostic accuracy. Studies in which the suspicious or pathologically proven metastatic axillary lymph node was marked only after NST, that is without clip placement before initiation of treatment, were excluded as this was not in agreement with the definition of TAD¹⁷. Studies that evaluated only excision of a TLN without SLNB were also excluded, as were those in which intraoperative localization of the TLN was not attempted (for example, only an X-ray was used to check whether the TLN was present in the surgical specimen). Studies that also included patients with clinically node-negative breast cancer or those who underwent primary surgery were excluded if it was not possible to identify the results specifically for patients with cN+ disease treated with NST. If more than one study reported on (part of) the same cohort, only that describing the largest cohort was included.

For qualitative synthesis, studies were grouped by type of definitive marker used and by timing of definitive marker placement. In one-step procedures, the definitive marker was placed before NST, followed by excision of the TLN during surgery. In two-step procedures, a clip was first placed before NST, followed by placement of a definitive marker adjacent to the clip after NST to enable subsequent excision of the TLN during surgery. In clinical practice, a wide variety of clips is used. When assessing the included studies, the specific type of clip used was not taken into account.

Identification of studies

PubMed and Embase were searched until 4 July 2023, without restriction on language or date of publication. The search strategies for both databases (Appendix S1) were checked by a librarian specialized in health sciences. The reference lists of included studies were checked for additional relevant studies, as were existing reviews.

Selection of studies

Reference management software (Endnote^® version 20.5, Philadelphia, PA, USA) was used to identify and remove duplicate references. The title and abstract of all remaining references, and subsequently the full text of potentially eligible studies, were evaluated independently by two authors. Disagreements regarding eligibility of studies were resolved in a consensus meeting.

Data extraction and analysis

The following variables were extracted from each included study: first author, year of publication, study design, sample size, percentage of patients with cN+ disease in whom nodal positivity at diagnosis was verified by pathology, type of tracer used for SLNB, type of definitive marker used for intraoperative excision of TLN, whether this marker was placed before or after NST, IR of the clipped TLN on imaging after NST (if applicable), IR of the TLN during surgery, percentage of patients who underwent ALND, proportion of SLNB and TLN being the same node (concordance), number of excised lymph nodes (mean or median), and whether immunohistochemistry was used for the assessment of excised lymph nodes. A second author was consulted in case of uncertainty.

The random-effects model for meta-analysis in the metaprop command in Stata^® SE16.1 (StataCorp, College Station, TX, USA) was employed to calculate the overall pooled estimate of the IR of the TLN during surgery for both one- and two-step procedures. Effect sizes with 95% confidence intervals and weights were provided in forest plots visualized by type of marker and for the whole group. The variability of IR estimates owing to heterogeneity among included studies was quantified using the I² index. The χ² test was used to assess statistical heterogeneity. The test was two-sided, and P < 0.050 was considered statistically significant.

Quality assessment

One author assessed the quality of the included studies, including the risk of bias, using the National Institutes of Health (NIH) Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies, which consists of 14 questions²⁹. All questions could be answered with yes, no, cannot determine, not applicable, or not reported. Based on these responses, studies were rated as having good, fair, or poor quality. A second author was consulted in the event of uncertainty.

Results

Study selection

The literature search identified 460 articles. After removal of duplicates, 277 titles and abstracts were screened, followed by full-text evaluation of 89 articles. Eventually, 51 studies with a total of 4512 patients were included in the qualitative synthesis (Fig. 1).

Fig. 1.

Flow chart showing selection of studies for review

NST, neoadjuvant systemic therapy; SLNB, sentinel lymph node biopsy.

Study characteristics

Characteristics of the included studies, sorted by type of definitive marker, are listed in Table S1. In 42 of 51 studies (82%), nodal positivity at diagnosis was proven by pathology in all patients. In 18 of 51 studies (35%), dual tracer (consisting of blue dye and radioisotope) was used routinely during SLNB. The percentage of patients who underwent ALND was available in 42 of 51 studies (82%), and varied from 22 to 100%. In 8 of 42 studies, all patients underwent ALND.

Type of definitive marker

Six definitive markers were used to mark the TLN, all in combination with SLNB. In 17 studies^{17,19,27,30–43}, on the day of surgery a wire was placed after NST in the clipped TLN. In 12 studies^{17,21,41,42,44–51}, the clipped TLN was marked with a ¹²⁵I seed, either before or after NST, and in 5 studies^52–56, a form of ^99mTc was used to localize and excise the clipped TLN. In three of five studies^52–54, the clipped TLN was injected with ^99mTc-labelled macroaggregated albumin under ultrasound guidance 1 day before surgery. In the other two^55,56, either ^99mTc-labelled Nanoscan tracer or ^99mTc-labelled nanocolloid was injected (peritumorally or periareolarly) to localize the SLN by single-photon emission CT (SPECT)/CT on the day of surgery or 1 day before, and to determine whether the clipped TLN was an SLN. If not, either ^99mTc-labelled Nanoscan tracer was injected into the clipped TLN, or a wire was placed under ultrasound guidance to enable excision of the clipped TLN. In both ¹²⁵I and ^99mTc marking, a hand-held γ probe was used to localize and subsequently excise the TLN during surgery. In 10 studies, the clipped TLN was marked with a magnetic marker^25,57–61, radiofrequency identification (RFID) tag⁵⁷, or an electromagnetic reflector^{43,57,62–64}, either before or after NST. At surgery, the TLN was localized using a hand-held probe based on magnetic fields, radiowave signalling, or radar/infrared technology respectively. In nine studies^26,65–72, the clipped TLN was tattooed with black ink (carbon, charcoal, or 4% carbon microparticle suspension), either before or after NST. Subsequently, it was excised under visual guidance during surgery. In two studies^73,74, the clipped TLN was localized and excised under intraoperative ultrasound (IOUS) guidance.

Timing of marker placement

Five studies assessed both one- and two-step procedures, whereas the remainder evaluated either a one- or two-step procedure. Tables 1 and 2 provide detailed information for one-step (15 studies) and two-step (41 studies) procedures respectively.

Table 1.

Studies describing a one-step procedure

Reference	Sample size	Type of definitive marker	IR at surgery (%)
Simons et al.42	68	125I seed	93
Rebollo Aguirre et al.48	6*	125I seed	97†
Simons et al.21	238	125I seed	94.1
Munck et al.51	142	125I seed	99.3
Martinez et al.60	44	Magnetic marker	44 of 44
Barry et al.61	54	Magnetic marker	98
Patel et al.66	47	Carbon ink	47 of 47
Natsiopoulos et al.26	75	Carbon ink	95
Allweis et al.67	63	Carbon ink	95
Dostalek et al.68	27	Carbon ink	22 of 27
de Boniface et al.69	149	Carbon ink	94.6
Pinto et al.70	13*	Carbon ink	8 of 13
Spautz et al.72	123	4% CMS	98.3
Pinto et al.73	37	Clip (IOUS)	30 of 37
Siso et al.74	235	Clip (IOUS)	96.2

^*Included as the total study comprised 25 patients or more. †Both one- and two-step procedures were assessed; an overall outcome was provided. IR, identification rate; CMS, carbon microparticle suspension; IOUS, intraoperative ultrasonography.

Table 2.

Studies describing a two-step procedure

Reference	Sample size	Type of definitive marker	IR on imaging after NST (%)	IR at surgery (%)
Plecha et al.30	73	Wire	n.r.	97
Dashevsky et al.31	28	Wire	28 of 28	26 of 28
Hartmann et al.32	30	Wire	24 of 30	17 of 24
Balasubramanian et al.27	25	Wire	25 of 25	23 of 25
Alarcon et al.33	28	Wire	28 of 28	28 of 28
Flores-Funes et al.34	60	Wire	97	97
Garcia-Novoa et al.35	42	Wire	42 of 42	42 of 42
Gurleyik et al.36	64	Wire	98	100
Sierra et al.37	51	Wire	n.r.	96
Kuemmel et al.19	423	Wire	c.d.	77.8*
Acea-Figueira et al.38	81	Wire	100	99
Sargent et al.39	62	Wire	n.r.	n.r.
Wu et al.40	239	Wire	c.d.	94.1*
Munck et al.41	543	Wire (263)	79.4†	90.1
		125I seed (103)		96.1
		Ink on skin (62)		82
		Magnetic marker (3)		3 of 3
Caudle et al.17	96	125I seed (94) Wire (2)	n.r.	n.r.
Diego et al.44	30	125I seed	29 of 30	29 of 29
Nguyen et al.45	25	125I seed	20 of 25	20 of 20
Beniey et al.46	35	125I seed	34 of 35	34 of 35
Simons et al.42	70	125I seed (12)	n.r.	11 of 12
		Wire (58)		95
Aragon-Sanchez et al.47	32	125I seed	29 of 32	31 of 32‡
Rebollo Aguirre et al.48	44	125I seed	n.r.	97§
Weiss et al.49	78	125I seed	c.d.	c.d.
Clark et al.50	77	125I seed	n.r.	97
Fuertes Manuel et al.52	30	99mTc	30 of 30	27 of 30
del Castillo et al.53	54	99mTc	n.r.	98
ella et al.54	77	99mTc	94	97
Winder et al.55	38	99mTc	n.r.	37 of 38
Dilege et al.56	61	99mTc	93	97
Laws et al.57	56	RFID tag (43)	95†	93†
		Magnetic marker (12)
		Electromagnetic reflector (1)
Sun et al.62	45	Electromagnetic reflector	n.r.	45 of 45
Balija et al.43	99	Electromagnetic reflector (57)¶	84	100¶
		Wire (42)	35 of 42	79†
Weinfurtner et al.63	105	Electromagnetic reflector	n.r.	100.0
Taj et al.64	80	Electromagnetic reflector	49	n.r.
Mariscal Martinez et al.58	30	Magnetic marker	30 of 30	30 of 30
Reitsamer et al.25	40#	Magnetic marker	40 of 40	40 of 40
Simons et al.59	51	Magnetic marker	98	100
Martinez et al.60	37	Magnetic marker	n.r.	37 of 37
Barry et al.61	74	Magnetic marker	98	76
Kim et al.65	28	Charcoal	n.r.	27 of 28
Pinto et al.70	18**	Carbon ink	n.r.	17 of 18
Porpiglia et al.71	32	Carbon ink	n.r.	27 of 32

^*An overall identification rate (IR) was provided (on imaging and during surgery combined). †More than one marking technique was assessed; an overall outcome was provided. ‡Three of 32 patients underwent stereotactic wire localization with mammography to enable excision. §Both one- and two-step procedures were assessed; an overall outcome was provided. ¶In 22 patients, the marker was placed in the clipped axillary lymph node before or during neoadjuvant systemic therapy (NST). #In two patients, the marker was placed directly before NST. **Included as the total study comprised 25 patients or more. n.r., Not reported; c.d., cannot determine.

Studies using a one-step procedure

Fifteen studies described a 1-step procedure, with a total of 1321 patients. In all studies, the definitive marker was placed in the metastatic or suspicious TLN before NST, followed by surgical excision after NST. The marking technique comprised the use of either a ¹²⁵I seed (4 studies), magnetic marker (2 studies), black ink (7 studies), or a clip combined with IOUS-guided localization (2 studies). Overall, the IR of the TLN at surgery varied from 8 of 13 to 47 of 47. When grouped by type of definitive marker, the IR ranged from 93.0 to 99.3, 98 to 44 of 44, 8 of 13 to 47 of 47, and 30 of 37 to 96.2% for ¹²⁵I seed, magnetic marker, black ink, and clip with IOUS-guided localization respectively. The overall pooled IR at surgery was 96 (95% c.i. 93 to 98)% (Fig. S1). Statistically significant heterogeneity was present between studies (I² = 73.2%, P < 0.001). The concordance rate between the TLN and SLN ranged between 47.9 and 100%.

Studies using a two-step procedure

Forty-one studies described a 2-step procedure, with a total of 3191 patients. In all studies, a clip was placed in the metastatic or suspicious TLN before NST. After NST, the clipped TLN was localized with imaging (ultrasonography in the vast majority), and was subsequently marked with either a wire (17 studies), ¹²⁵I seed (10 studies), ^99mTc (5 studies), (electro)magnetic/radiofrequency marker (11 studies), or black ink (3 studies). The IR of the clipped TLN on imaging was reported in 23 of 41 studies, ranging from 49 to 100%. In 18 of 41 studies, the IR of the clipped TLN could not be determined on imaging (only an overall IR was provided in 3 studies), or it was not reported (15 studies; mostly because patients were excluded from analyses in the event of unsuccessful localization of the clipped TLN on imaging). Overall, the IR of the TLN at surgery varied from 17 of 24 to 100%. When grouped by type of marker, the IR at surgery ranged from 17 of 24 to 100, 11 of 12 to 29 of 29, 27 of 30 to 98, 76 to 100, and 82 to 27 of 28% respectively for wire, ¹²⁵I seed, ^99mTc, (electro)magnetic/radiofrequency markers, and black ink. The IR at surgery could either not be determined or was not reported in six studies. The overall pooled IR was 97 (95% c.i. 95 to 98)%. Statistically significant heterogeneity was present between studies (I² = 69.3%, P < 0.001) (Fig. S2). The concordance rate between the TLN and SLN was reported in 28 studies and ranged from 35.7 to 91.0%.

Quality assessment

Assessed using the NIH Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies, 8 studies were rated as being of good quality, 40 of fair quality, and 3 of poor quality (Table S2).

Discussion

Worldwide, several different surgical procedures are being used in clinical practice for axillary staging after NST in cN+ breast cancer. Most institutions now prefer less invasive staging procedures, including SLNB alone, excision of a TLN, or the TAD procedure, with the aim of enabling response-guided axillary treatment after NST^75–78. This systematic review included 51 studies of TAD with a total of 4512 patients, and a wide range of TLN excision techniques were identified. Six definitive markers were recognized: wire, ¹²⁵I seed, ^99mTc, (electro)magnetic/radiofrequency markers, black ink, and clips (with IOUS-guided localization and excision). Apart from this, variations in timing of definitive marker placement were assessed.

The use of wire-guided localization is both accessible and inexpensive⁷⁹. The wire, however, needs to be placed 1 day before operation or on the day of surgery, which requires adequate planning. Furthermore, the wire may dislocate in the event of patient movement or manipulation during surgery, which can complicate retrieval of the clipped TLN³². The wire may be also be uncomfortable for the patient. A ¹²⁵I seed does not have to be placed on the day of surgery, and can even be placed before NST. In addition, the use of a hand-held γ probe facilitates identification of the TLN¹⁶. A downside is that the use of ¹²⁵I seeds is strictly regulated, making widespread application difficult because many countries do not allow them to be used for diagnostic purposes, or only allow them if the ¹²⁵I seed is placed after NST⁷⁹. An alternative would be to mark the TLN with ^99mTc, which is inexpensive, already widely applied for diagnostic purposes, and the use of a hand-held γ probe facilitates localization of the TLN during surgery^52–54. A downside is its short half-life of 6 h, so it has to be injected just before surgery^52–54. If ^99mTc is not injected into the TLN itself, but peritumorally or periareolarly (as is already part of routine SLNB), and the clipped TLN is an SLN on SPECT/CT, an additional procedure, for example injecting ^99mTc-labelled Nanoscan tracer into the clipped non-SLN to enable excision, is not needed. Magnetic markers, RFID tags, and electromagnetic reflectors are promising non-radioactive alternatives, which can all be placed before the start of NST, and are localized with a hand-held probe to facilitate intraoperative excision of the TLN^{25,43,57–64}. In the case of the RFID tag and electromagnetic reflector, the probe also displays the distance from the tip of the probe to the marker⁸⁰. As these three markers are not radioactive, there are no regulatory issues, but they are more expensive and require purchase of additional instruments, such as the localization device⁷⁹. In addition, the magnetic marker and RFID tag both create an artefact on MRI^25,81, complicating response evaluation, especially when the primary tumour is located in the lateral upper quadrant. Employing a magnetic marker also requires use of non-magnetic equipment during surgery. The electromagnetic marker may also create minimal artefacts⁸¹. Currently, the magnetic marker is being updated, in an effort to reduce MRI artefacts and to avoid the need for non-magnetic equipment⁸². Another non-radioactive and inexpensive technique is to tattoo the TLN with black ink. As this technique lacks a detection probe and the ink cannot be visualized on imaging, it is more difficult to localize the TLN during surgery, and the IR for this type of marker was reported to be as low as 61.5%. Moreover, studies^26,79 have described spontaneous migration of black ink, but also deliberate distribution of black ink around the TLN to increase the IR^65,66. In both instances, this can result in unnecessary excision of additional lymph nodes^26,66, increasing the risk of postoperative morbidity. Finally, IOUS-guided excision of the clipped TLN is possible, which is inexpensive and does not require additional markers or the purchase of new instruments. It does require an ultrasound machine in the operating room, and a specialist qualified to perform IOUS⁷⁹.

As a result of the abovementioned benefits and drawbacks of the different techniques, institutions and/or specialists each have their own TAD preferences, resulting in a wide variety of techniques used in daily practice. As the included studies are very heterogeneous with a broad range of reported IRs, it is not possible to conclude which technique is superior in identifying the TLN. This systematic review, however, does show an important drawback of two-step procedures that breast cancer specialists need to take into consideration. The TLN needs to be localized twice, not only at surgery, but also after NST in order to place the definitive marker. The ability to localize the clipped TLN on imaging after NST varied from 49 to 100%. Importantly, 18 of 41 studies did not report any data regarding localization of the clipped TLN. The wide variation in ability to localize the clipped TLN on imaging may be explained by the diverse range of clips used in clinical practice. In addition, it may be influenced by the level of experience of the specialist performing the localization, and whether or not this is done by a dedicated breast cancer specialist. Furthermore, the inability to identify the TLN on imaging after NST is possibly explained to the fact that the visibility of clips decreases with time⁸³. When a hyperechogenic clip is placed in the hypoechogenic cortex, regression of the cortex in the event of response to NST can also affect the visibility of the clip or cause the clip to dislocate⁸⁴. This is in accordance with the multivariable analyses of Kuemmel et al.¹⁹, in which an axillary pCR on imaging was also associated with inability to identify the TLN at surgery. Hence, it is important to use a clip with good visibility on ultrasonography.

A large number of studies describing experiences with marking techniques for TLN excision were identified in this systematic review. Although it is of great importance that these studies are performed to share experiences, the included studies also had some limitations. Most had a relatively small sample size, with study populations ranging from 25 to 543 patients. Twenty-four studies had fewer than 60 patients. For example, in the study of Pinto et al.⁷⁰, which assessed both one- and two-step procedures with carbon ink in a prospective cohort, the IR of the TLN at surgery was 61.5% for the one-step procedure. This was, however, based on a small subgroup of the study population (8 of 13 patients). Another limitation was the retrospective (45% of studies) or single-centre (80%) study design. Moreover, the definition of IR was not always clear and, for two-step procedures, the IR of the clipped TLN on imaging was not provided in 18 of 41 studies. Because of these limitations and study heterogeneity, the results of the random-effects model should be interpreted with caution. Finally, it was not considered whether, at the time of diagnosis, the definitive marker (in a 1-step procedure) or clip (in a 2-step procedure) was placed directly after fine-needle aspiration cytology or core needle biopsy of the suspicious axillary lymph node, or if this was done after the lymph node had been shown to be metastatic by pathology. Along this line, the assessment did not include the different types of clip used for marking the TLN before NST, which also likely varies between, and even within, institutions.

High-quality prospective studies are thus needed that evaluate both one- and two-step procedures, provide a clear definition of IR, and take into account the results of clip identification on imaging in two-step procedures. Currently, the Magellan trial (NCT03796559) is recruiting patients in a prospective study evaluating a magnetic marker in a one-step procedure. In addition, Hartmann et al.⁸⁵ recently published results regarding the applicability of a magnetic marker in one-step procedure in a multicentre cohort of 151 patients. The TLN was removed successfully in 146 patients, which resulted in an IR of 96.0%. Response assessment with MRI was reported to be compromised in 15 of 151 patients (9.9%). Furthermore, in the prospective IMTAD study⁸⁶, which included 189 patients, marking with a ¹²⁵I seed (135 patients), magnetic marker (30), or carbon suspension (24) after NST in a clipped TLN are being compared. Recently published results demonstrated comparable complication rates regarding marker placement and localization, and marker dislodgement.

In the meantime, while TAD and other less invasive axillary staging procedures are being performed in daily practice worldwide, limited but increasing evidence is available regarding the oncological outcomes of response-guided treatment based on less invasive axillary staging procedures. Interestingly, although these procedures were initially introduced to omit ALND in the event of an axillary pCR, ALND is now also being omitted in selected patients with residual disease⁷⁵. Van Loevezijn et al.²³ recently published 3-year follow-up results of the MARI protocol, in which axillary treatment decisions were made based on the findings of [¹⁸F]fluorodeoxyglucose PET–CT in combination with the outcome of the MARI procedure. ALND was omitted in 217 of 272 patients (80.0%) and replaced by axillary radiotherapy in 161 (74.2%) in this single-centre study, with a 3-year axillary recurrence-free survival rate of 98.0 (95% c.i. 96.0 to 100)%. NSABP-B51/RTOG 1304 and ATNEC (NCT01872975 and NCT04109079 respectively) are ongoing RCTs evaluating ALND and/or locoregional radiotherapy in patients with cN+ breast cancer treated with NST, and are including patients with ypN0 disease, whereas Alliance A011202 and TAXIS (NCT01901094 and NCT03513614 respectively) are including patients with ypN+ disease. Together with registry studies such as MINIMAX and AXSANA^83,87, these trials will provide more evidence about appropriate locoregional treatment strategies for cN+ disease in terms of long-term prognosis, in order to prevent overtreatment as well as undertreatment. In addition, these trials may help determine the optimal procedure for axillary staging in such patients, not only in terms of IR and feasibility but also oncological safety and quality of life. With regard to quality of life, the number of excised lymph nodes should also be taken into account, as this can affect arm morbidity. For instance, excision of three or more SLNs may be required when SLNB alone is performed (to improve the FNR), whereas TAD may involve the removal of a single lymph node.

The present systematic review has underlined the scarcity of high-quality studies, rendering it impossible to determine the optimal procedure in terms of IR and feasibility. Each TLN excision technique, however, has its own benefits and drawbacks that should be taken into consideration when performing TAD in clinical practice.

Funding

S.R.d.W. received a salary from the Dutch Cancer Society (KWF - Pink Ribbon, grant number 12518). The funder was not involved in the study design, data collection, data analysis, manuscript preparation or publication decisions.

Author contributions

Sabine De Wild (Conceptualization, Investigation, Methodology, Writing—original draft, Writing—review & editing), Linetta Koppert (Supervision, Writing—review & editing), Thiemo van Nijnatten (Writing—review & editing), Loes Kooreman (Writing—review & editing), Marie-Jeanne Vrancken Peeters (Supervision, Writing—review & editing), Marjolein Smidt (Supervision, Writing—review & editing), and Janine Simons (Conceptualization, Investigation, Methodology, Supervision, Writing—review & editing)

Disclosure

The authors declare no conflict of interest.

Supplementary material

Supplementary material is available at BJS online.

Data availability

No new data were generated or analysed in this manuscript.